I've been checking out Pixels lately, and my mindset is pretty twisted: on one hand, I've seen too many old tricks in Web3 games like 'subsidy - farm - crash', where the ROI always ends up in the red; on the other hand, the narrative of Pixels doesn't feel like storytelling — it’s more like treating 'rewards' as a measurable, optimized, and reusable growth system. The key change they keep emphasizing is Stacked: it's not just a regular rewards app, but a 'rewarded LiveOps engine', layered with an AI game economist that uses behavioral data to decide 'who, when, how much, and why to give'. It sounds a bit like operational jargon, but they lay out the verifiable anchors directly: processed over 200M+ rewards, generated 25M+ in revenue, and even boast about '1M DAU' scale.

I want to talk about business ROI, so let's skip the price chatter—those price fluctuations are just noise in the face of such projects (and going too deep into it can lead us off track). What I'm really concerned about is: if I treat Pixels as a company that 'views rewards as investments,' what exactly drives its ROI? Is there a chance this could shift from 'Pixels being used in-house' to a B2B engine that other games are willing to pay for? If the answer is 'possible,' then its business model isn't the old P2E playbook, but something closer to 'LiveOps infrastructure + data flywheel.'

First, let's clarify this ROI thing: traditional Web3 games die in two pits. The first pit is treating rewards as a 'cost center,' giving them away without a second thought, which can even hurt the economy (rewards being snatched by bots, concentrated sell-offs from farms, squeezing the experience for real players, resulting in worse retention). The second pit is the non-attributable ROI—you simply don’t know how much 'new retention/new revenue' comes from an event, and you can only fool yourself with 'hype' and 'community sentiment.' What Pixels/Stacked aims to solve is precisely these two pits: rewards are no longer just indiscriminate airdrops, but attributable 'investments,' and they use data to zero in on 'incremental gains.'

I explain it in a more 'operational dog' way: ROI = (incremental revenue + future revenue from incremental retention) / (reward cost + operational cost + cheating loss). In the past, most projects would only focus on 'reward cost,' thinking that cutting back a bit would be safer, resulting in quicker retention drops; or they’d only focus on 'incremental revenue,' running a big event to spike the curve temporarily, only to have players leave more thoroughly afterward. The real way to make ROI work is to break down 'incremental retention' at the cohort level—D1, D7, D30 retention, reflow rates, payment conversion, ARPPU, LTV—and treat reward deployment as an experimental variable, using A/B testing or quasi-experimental approaches (like stratified randomization, geographic/channel slicing) for attribution. If that 'AI game economist' on Stacked can truly achieve 'understanding player behavior—providing experimental recommendations—implementing reward logic—then recovering results,' it’s not just a gimmick but pushes LiveOps from 'empiricism' to 'dataism.'

But I'm not excited just because of the term 'AI'—I'm actually more cautious. Because AI can easily turn into two things in operations: one is a 'pretty report generator,' written like a consultancy, but relies entirely on human effort; the other is an 'overfitted deployment machine,' making metrics look good in the short term while damaging the ecosystem in the long run. For Pixels' setup to work, it must meet a hard condition: it has to be refined in a 'real production environment,' facing cheating, dealing with rewards being exploited, and managing player fatigue from events, and still be able to iterate stably. I think they’ve hit the right narrative here—repeatedly emphasizing it’s grown from Pixels' own long-cycle operations, not just some white paper fantasy; plus, they anchor results around 'processing 200M+ rewards, contributing 25M+ revenue.'

Now, let me break down ROI from the 'where does the money come from' perspective. Many people reflexively ask when they see '25M revenue': is it from tokens? Is it from NFTs? But one narrative explicitly states 'real in-game purchases, VIP subscriptions, cosmetics, upgrades, passes,' etc., rather than relying on 'tokens to pump the price.' If this structural part holds true, it’s actually closer to traditional game revenue models, just treating Web3's assets/reward mechanisms as tools to 'improve retention and payment efficiency,' not as revenue itself.

This point is crucial: when revenue comes from real player payment behavior, ROI has stable numerators and denominators; otherwise, you’ll fall back into the old trap of 'using subsidies to create false prosperity.'

So how do we control reward costs? This is the 'line between life and death' for P2E. My personal judgment is that if Stacked wants to win on ROI, the core isn't 'giving less,' but 'giving smartly': upgrading rewards from 'task-based distribution' to 'behavior quality distribution.' For example, take the same task completion: bots can spam frequently, but their behavior sequence will be very 'clean' (single path, stable time intervals, shallow interactions); real players' behaviors are messier (lingering, hesitating, socializing, exploring, switching systems). If the system can use these characteristics to make 'differentiated pricing for rewards,' you'll see a direct result: reduced reward waste rate and increased incentive efficiency for real players. You could even make 'cheating loss' a separate ROI deduction item, requiring every event to provide a 'suspected cheating percentage, reward recovery/frozen ratio, and abnormal path hit rate' follow-up. Stacked repeatedly emphasizes 'determining rewards based on real player behavior,' which is essentially moving in this direction.

I want to add something slightly 'counterintuitive': many projects view anti-cheating as a cost, but projects like Pixels might actually turn anti-cheating into a moat. The reason is quite straightforward—when you treat rewards as a deployment budget, cheating directly steals from that budget; the better you are at identifying real players, the higher your ROI can go, and the more you dare to increase investments; the more you dare to increase investments, the more data you gather, and your ability to identify cheating and player behavior strengthens. Once this feedback loop kicks in, it's hard for newcomers to catch up by just 'copying gameplay,' because they lack the long-cycle behavior data and counter-experience, not just the event copywriting.

Breaking it down further: if Pixels is just using Stacked for its own games, even if the ROI is great, it's merely about internal efficiency. The ceiling depends on how many games Pixels can create and how many players it can attract. What really catches my eye is its push towards being a 'B2B infrastructure': Stacked targets both players for tasks and rewards, and game developers for LiveOps engines and AI insights.

But I also have to throw a bit of cold water on it: the B2B ROI logic is more stringent—why would external studios want to use you? Can using you measurably improve their retention/payment? Can it help them avoid pitfalls? Can it save operational manpower? If Stacked can’t deliver 'third-party client cases + replicable methodologies,' then it risks staying at the level of 'Pixels is good at running its own games,' becoming just a pretty story shell. Some even bluntly point out: if it remains private infrastructure, not adopted by other games, then so-called B2B is just a thin veneer.

So when I look at Pixels' commercial ROI, I'm more focused on one metric: not price, not hype, but whether 'Stacked really delivers incremental effects for third parties.' Once that emerges, you'll see a chain reaction: more event budgets willing to flow in, more data coming back to train models, more gameplay being modularized, and only then could we form 'infrastructure profits.'

Now we should talk about PIXEL's position in ROI (I’ll try to be restrained because you probably don’t want me to write a market analysis). I prefer to view PIXEL as 'fuel for cross-game rewards/loyalty currency,' rather than a 'single-game token.' Some content explicitly uses similar expressions: not just a token for one game, but a vehicle for rewards and loyalty across the ecosystem.

From the ROI perspective, the significance of this design is that when you upgrade rewards from 'single-game subsidies' to 'cross-game entitlements,' you can reduce the risk of single-point economics being drained—rewards can be channeled, reclaimed, and priced at different tiers, while also solidifying player identities and entitlements into long-term assets. Note that I’m talking about 'possibilities,' not promises; whether this holds true ultimately depends on whether Stacked can onboard more games and whether those games are genuinely willing to use PIXEL as part of their reward settlement or entitlement medium (even if just partially).

If I were the operations lead, how would I use the ROI framework to validate Pixels/Stacked? I'd focus on three 'lifeline metrics' that are unrelated to price but can directly reflect whether the business model is healthy. The first is the attributable ROI of events: can we provide 'control group vs. experimental group' incremental retention, incremental revenue, and reward waste rate for every big reward event? If it's just 'a surge in participants,' that's self-indulgence. The second is the cheating cost curve: does the cheating percentage rise or fall as reward scales increase? If the system grows and cheating becomes harder, its moat is real; if it gets exploited right away, no matter how smart the AI is, it's just a PowerPoint presentation. The third is the B2B diffusion speed: are there new games coming on board? After onboarding, are there publicly available 'before-and-after comparison metrics' or at least verifiable cooperation facts? Without external clients, the ROI ceiling is locked within Pixels.

I know many will say: 'Aren't you just turning Web3 games into traditional growth analysis?' Yes, that's exactly my point. Because I increasingly feel that the small fraction of Web3 games that survive won't be defined by narrative but by systematic operational abilities—treating rewards as investments, treating economics as engineering, and treating players as long-term relationships rather than one-off traffic. By pushing Stacked to the forefront, Pixels is actually admitting something that the industry is reluctant to acknowledge: more rewards aren't necessarily better; rewards should be audited, optimized, and held accountable just like a marketing budget.

Lastly, I must return to my own attitude: I won’t automatically believe everything just because of '200M rewards, 25M revenue,' but I'm willing to treat it as a solid starting point—at least it indicates that this system has been tested at a real user scale.

Next, it has to prove it's 'replicable': from Pixels' in-house use to third-party adoption, from improving single-ecosystem ROI to diffusing multi-ecosystem ROI. If it can't pass this hurdle, it's just a strong internal operations system; if it does, it could become a rare 'infrastructure-level business' in Web3 gaming. I write this without pretending to be certain—I’ll keep my eyes on the 'third-party clients and incremental effects' line, considering all other noise as just that.

\u003cm-23/\u003e ,\u003cc-25/\u003e, \u003ct-27/\u003e